Maintenance Notice

Due to necessary scheduled maintenance, the JMIR Publications website will be unavailable from Wednesday, July 01, 2020 at 8:00 PM to 10:00 PM EST. We apologize in advance for any inconvenience this may cause you.

Who will be affected?

Accepted for/Published in: Journal of Medical Internet Research

Date Submitted: Nov 21, 2024
Date Accepted: Feb 7, 2025

The final, peer-reviewed published version of this preprint can be found here:

Diagnostic Performance of Artificial Intelligence–Based Methods for Tuberculosis Detection: Systematic Review

Hansun S, Argha A, Bakhshayeshi I, Wicaksana A, Alinejad-Rokny H, Fox G, Liaw ST, Celler BG, Marks GB

Diagnostic Performance of Artificial Intelligence–Based Methods for Tuberculosis Detection: Systematic Review

J Med Internet Res 2025;27:e69068

DOI: 10.2196/69068

PMID: 40053773

PMCID: 11928776

Diagnostic Performance of Artificial Intelligence-Based Methods for Tuberculosis Detection: Systematic Review

  • Seng Hansun; 
  • Ahmadreza Argha; 
  • Ivan Bakhshayeshi; 
  • Arya Wicaksana; 
  • Hamid Alinejad-Rokny; 
  • Gregory Fox; 
  • Siaw-Teng Liaw; 
  • Branko G Celler; 
  • Guy B Marks

ABSTRACT

Background:

Tuberculosis (TB) remains a significant global health concern, contributing to the highest mortality among infectious diseases worldwide. In response, various Artificial Intelligence (AI)-based methods have been developed to address this issue.

Objective:

This review aims to comprehensively evaluate the performance of AI-based algorithms in TB detection.

Methods:

Following the Preferred Reporting Items for Systematic Reviews and Meta-Analysis (PRISMA) 2020 Guidelines, we conducted a systematic review to synthesize current knowledge on this topic. Our search across three major databases (Scopus, PubMed, ACM Digital Library) yielded 1,146 records, from which we independently screened, reviewed, and assessed studies for inclusion. Ultimately, 152 studies met our inclusion criteria. The Quality Assessment of Diagnostic Accuracy Studies version 2 (QUADAS-2) was performed for the risk of bias assessment of all included studies.

Results:

Radiographic biomarkers and Deep Learning (DL) approaches were predominantly utilized, with Convolutional Neural Networks (CNNs) employing VGG-16 (n=37), ResNet-50 (n=33), and DenseNet-121 (n=19) architectures being the most common DL approaches. AI methods demonstrated very good performance against a variety of reference standards, achieving a mean accuracy of 91.93% (median 93.59%), mean AUC of 93.48% (median 95.28%), mean sensitivity of 92.77% (median 94.05%), and mean specificity of 92.39% (median 95.38%) for all included studies. Notably, only one study conducted domain-shift analysis for TB detection.

Conclusions:

Findings from this review underscores the considerable promise of AI-based methods in the realm of TB detection. Future research endeavors should prioritize conducting domain-shift analyses to better simulate real-world scenarios in TB detection. Clinical Trial: PROSPERO CRD42023453611


 Citation

Please cite as:

Hansun S, Argha A, Bakhshayeshi I, Wicaksana A, Alinejad-Rokny H, Fox G, Liaw ST, Celler BG, Marks GB

Diagnostic Performance of Artificial Intelligence–Based Methods for Tuberculosis Detection: Systematic Review

J Med Internet Res 2025;27:e69068

DOI: 10.2196/69068

PMID: 40053773

PMCID: 11928776

Download PDF


Request queued. Please wait while the file is being generated. It may take some time.

© The authors. All rights reserved. This is a privileged document currently under peer-review/community review (or an accepted/rejected manuscript). Authors have provided JMIR Publications with an exclusive license to publish this preprint on it's website for review and ahead-of-print citation purposes only. While the final peer-reviewed paper may be licensed under a cc-by license on publication, at this stage authors and publisher expressively prohibit redistribution of this draft paper other than for review purposes.